A hazards approach towards modelling pandora moth risk

A hazards approach towards modelling pandoramoth riskJames H. Speer* and Ryan R. Jensen Department of Geography, Geology, and Anthropo-

logy, Indiana State University, Terre Haute, IN, USA

Abstract

Aim Pandora moth (Coloradia pandora Blake) is a phytophagous insect that produces adistinctive tree-ring pattern in ponderosa pine (Pinus ponderosa Dougl. ex. Laws.)during outbreak cycles. This paper describes the spatial characteristics of the outbreakregions, determines whether the size of the 1989 outbreak was within the historicalrange of variability, and constructs a hazard map identifying the forests in Oregon thatare susceptible to future pandora moth outbreaks.

Location South-central Oregon along the eastern flank of the Cascade mountain rangein the High Lava Plains and Basin and Range Provinces.

Methods We used dendrochronological records of 17 pandora moth outbreaks on14 sites over 31,200 km2 area spanning 433 years. Using the site locations, we calcu-lated minimum bounding polygons of adjacent recording sites to determine the relativesize of each outbreak. Published literature on past pandora moth outbreaks and theenvironmental conditions of locally known outbreaks were used to create an outbreakhazard map using a geographical information system (GIS) model. Vegetation, climate,and soil layers were used to determine the potential susceptibility of Oregon forests topandora moth.

Results We found the area affected by past pandora moth outbreaks ranged in size from12.4 to 3,391.5 km2. The 1989 outbreak covered 807.9 km2, which was well within thehistorical range of variability. The vegetation and soil layers greatly restricted the areasusceptible to pandora moth while the climate layer seemed to have little effect inrestricting the susceptible area.

Main conclusions Pandora moth outbreaks did not increase in size over the last centuryas we have seen with spruce budworm outbreaks in this same region. Analysis of theenvironmental variables that are known to affect pandora moth outbreaks enabled us toproduce a hazard map that predicts the suitable habitat for pandora moth. Temperatureat the landscape scale did not restrain the range of pandora moth. The GIS modelenabled us to propose areas susceptible to future pandora moth outbreaks providing apredictive model that can now be tested and refined with further sampling.

Keywords

Pandora moth, Coloradia pandora, dendrochronology, dendroecology, ponderosa pine,Oregon, geographical information system, hazard map, insect outbreak.

INTRODUCTION

Pandora moth (Coloradia pandora Blake, Lepidoptera:Saturniidae) is a phytophagous insect endemic to the United

States whose range includes most western states exceptIdaho and Washington [most notably in Arizona, California,Colorado and Oregon (Carolin & Knopf, 1968)]. In Oregonthe main host species is ponderosa pine (Pinus ponderosaDougl. ex Laws.) and occasionally lodgepole pine (Pinuscontorta Dougl. ex Loud.). Pandora moth has been littlestudied, but recent interest has arisen due to an outbreak incentral Oregon from 1988 to 1996 (Speer, 1997).

*Correspondence: James H. Speer, Department of Geography, Geology, and

Anthropology, Indiana State University, Terre Haute, IN 47809, USA.

E-mail: [email protected]

Journal of Biogeography, 30, 1899–1906

� 2003 Blackwell Publishing Ltd

In a previous study, Speer et al. (2001) used tree rings toreconstruct 22 pandora moth outbreaks over a 622-yearperiod from 14 sites throughout central Oregon (Fig. 1).Speer et al. (2001) demonstrated that pandora moth out-breaks could be accurately reconstructed and identified aclear tree-ring pattern associated with the outbreaks. In thecurrent study, we use this reconstruction of past pandoramoth outbreaks to examine the spatial characteristics ofindividual outbreaks, examine the regional distribution ofoutbreaks through time, and parameterize a model devel-oped in a geographical information system (GIS) to deter-mine the potential habitat for pandora moth. The resultinghazard map can be used by managers to plan for futureoutbreaks and also as a guide to locate samples for futurepandora moth reconstructions. Thus, it may be used in forestmanagement planning as well as a sampling tool for futureresearch.

GIS has been used extensively in ecological studies toexamine spatial phenomena such as longleaf pine sandhillloss in the south-eastern United States coastal plain (Jensen& Carson, 2001) and forest fragmentation in southernVirginia, USA (Wickham et al., 2000). Specifically, GIS hasbeen used to study historical ecological events. For example,Jensen (2002) used GIS to examine temporal leaf area indexdynamics in north-central Florida, USA from 1972 to 1997.Baker & Kipfmueller (2001) used GIS to construct andanalyse the spatial ecology of pre-Euro-American fires in asouthern rocky mountain subalpine forest landscape. Eeleyet al. (1999) used GIS to examine the influence of climatechange on forest distribution in South Africa, and Baker(1995) used GIS to simulate the effects of the density oflandscape patches to long-term disturbance regimes – globalwarming and cooling and fragmentation – for a 400-yearperiod. Although, GIS has been used extensively in many

ecological studies our application is unique because we useGIS to model insect outbreak hazard and to map the area ofpast insect outbreaks reconstructed from tree-ring records.

Pandora moth

Pandora moth has a 2-year life cycle and spends their firstwinter in the second instar stage at the base of pine needles.At the end of their second summer the larvae move down thestem of the tree and burrow into the soil to pupate (Massey,1940; Carolin & Knopf, 1968; Schmid & Bennett, 1988;Fitzgerald, 1992). This insect is also closely adapted to itshost species only causing 2% mortality although defoliationis complete every other year. Because of the dependence ofthis insect on its host tree species, climate factors, and soiltype we use these parameters to map the area of Oregon thatis susceptible to pandora moth.

Temperature affects the survival of pandora moth,because the larvae are exposed to weather for one wholeyear. The larvae cease to feed at temperatures below 15.6 �C(Massey, 1940; Schmid & Bennett, 1988), and temperaturesat )40 �C have been shown to kill many caterpillars duringtheir first winter (Massey, 1940). Aspect (and most likely therelated temperature differences) has been shown to affectpandora moth distributions in Colorado. Massey (1940)found that pandora moth cause greater damage on south-east facing slopes than on north-west facing slopes.

Soil conditions are also a factor in pandora moth out-breaks. Pandora moths occur in areas with immature soilsoften developed from weathered granite or ash and pumicedeposits, which are loose enough to permit the larvae toburrow (Patterson, 1929; Massey, 1940; Carolin & Knopf,1968; Furniss & Carolin, 1977). In Oregon, pandora mothprefers soils associated with ash and pumice deposits from

Study site Figure 1 Hill-shaded map of Oregon withdots designating the location of the 14 study

sites in south-central Oregon.

� 2003 Blackwell Publishing Ltd, Journal of Biogeography, 30, 1899–1906

1900 J. H. Speer and R. R. Jensen

Mt Mazama, which erupted c. 6000 years ago (Patterson,1929; Massey, 1940). Massey (1940) observed that in theabsence of such soils, the larvae spend a considerableamount of time trying to burrow searching for a suitableplace for pupation. Larvae that make several attempts toborrow into soil that was above 50 �C eventually died fromexhaustion and exposure. In areas lacking the proper soil, itis likely that the outbreaks would not be able to maintainsufficient momentum to carry on the outbreak.

In Oregon, ponderosa pine is the main host species topandora moth, but lodgepole pine is also affected. Theponderosa pine forests of Oregon were heavily logged in theearly 1900s, effectively removing much of the moth’s foodsource. These forests have regrown as ponderosa pine/lodgepole pine forests (Franklin & Dyrness, 1988). In theBlue Mountains of Oregon, the pine trees have frequentlybeen logged and surface fires have been suppressed, resultingin an increase in spruce/fir forests (Mutch et al., 1993). Thisvegetation change has increased the habitat available tospruce budworm resulting in larger and more damagingbudworm outbreaks in the twentieth century (Swetnamet al., 1995). This study will address the issue of whetherpandora moth has been similarly affected by 20th centurylogging through the removal of ponderosa pine trees, whichare pandora moth’s main host species. Theoretically, pan-dora moth outbreaks would be smaller in extent and severitybecause of removal of a preferred host species.

METHODS

Site description

The study area is located on the east side of the CascadeMountains in south-central Oregon (43.10 N, 121.75 W). Itencompasses 31,200 km2 (240 km north-to-south and130 km east-to-west) (Fig. 1) and ranges in elevation from1320 to 1670 m. It includes most of the known range ofpandora moth in Oregon that was defined from historicaloutbreaks (Carolin & Knopf, 1968). Each of the 14 insectoutbreak sites was a disjunct stand of old-growth ponderosapine, covering 5–10 ha. The sites are located where remnantstands of old ponderosa pine trees were not logged, thereforethe spatial distribution of the sample sites were predeter-mined by past human land-use.

Determining the area of past outbreaks

Speer et al. (2001) reconstructed 22 outbreaks over the past622 years (from AD 1373 to AD 1995) by examining incre-ment cores collected from 140 mature ponderosa pine trees.Sample depth decreased further back in time reducing thenumber of recording sites. In this study, we truncate theirrecord at AD 1563 because the lack of recording sites in theearly part of the chronology would bias the site distribution.Six of the 14 sites were still recording outbreaks during the1563 outbreak.

Speer et al. (2001) demonstrated that the outbreak epi-sodes were not synchronous across the landscape, but

instead seemed to spread from one site to another. Because ofthis starting date ambiguity, we defined the date of the out-break episode as the period during which the outbreak wasmost active across the landscape. We used a number of cri-teria to determine when most of the sample area wasexperiencing an outbreak. From all of the sites recording eachoutbreak, we calculated the mode of the first year, the modeof the smallest ring, and the mode of the last year of the tree-ring pattern. When the mode of the first year was followed bythe mode of the smallest ring (1 or 2 years later), we wereconfident that the beginning date of the outbreak was relat-ively well defined. We used the mode of the last year of theoutbreak pattern as the end date of the outbreak. When theend dates did not coincide, we chose an end date that was inthe middle of the temporal distribution and did not overlapwith the beginning date of a subsequent outbreak.

Nearest neighbour analysis was performed to determinethe randomness of the sites affected by each outbreak. Thiswas calculated by comparing the mean of the actualdistances between each affected site in each outbreak withthe mean of the expected distances if the distribution of siteswas random. The difference between these two means wasthen divided by the standard error of the mean nearest-neighbour distance and compared with a critical value(Z ¼ 1.96; a ¼ 0.05; Earickson & Harlin, 1994).

GIS analysis

The locations of the sites were digitized into ArcView GIS3.2 to determine the minimum-bounding polygon encom-passing pandora moth outbreaks. Individual latitude andlongitude values were typed into the GIS to digitize thepolygons. These polygons were compiled by including allcontiguous sites that were affected by outbreaks. If anintervening site did not record an outbreak, then that poly-gon was interrupted and separate polygons were used tooutline the outbreak regions. For outbreaks that onlyaffected one site, we assumed a circle with a 2.5-km radius.This radius was chosen to account for the spread of pandoramoth to adjacent forest stands and approximates the size ofthe smallest historically recorded outbreak. From theseminimum-bounding polygons we estimated the size of pastoutbreaks. These calculations gave us a conservative esti-mate of the area affected by past outbreaks. To remainanalogous, we calculated the size of the 1989 outbreak usingthe above method, and then compared it with the size of pastoutbreaks to determine if the current outbreak was withinthe historical range of variability. The historical range ofvariability is defined by the documented past occurrence ofsome event and may be used to establish acceptable limits ofecosystem change (Morgan et al., 1994).

Hazard mapping of susceptibility to pandora moth

outbreaks

We determined the preferred condition of pandora mothoutbreaks by using our sample sites with known outbreakhistories to identify the soils and vegetation that make a


Hazard of pandora moth outbreaks 1901

site susceptible to pandora moth outbreaks. GIS vector datasets of soil (United States Geological Survey (USGS) soilsmap at the suborder level), vegetation (derived fromAdvanced Very High Resolution Radiometer (AVHRR)remotely sensed imagery), and climate (National WeatherService) were combined to estimate risk of potential out-break. Past pandora moth outbreaks in this area werelocated on Orthent, Cambid, Cryept and Xerept soils, sowe used these suborders as parameters in the hazard map(Table 1). We used the vegetation layer designated asponderosa pine/lodgepole pine from the AVHRR vegetationclassification to map the host tree species of this moth. Forthe temperature constraint we used the criterion that theaverage temperature had to exceed 15.6 �C for 90 days.This time period was determined based on the assumptionthat the larvae would need 3 months of intensive feeding tobe able to complete its life cycle. Average monthly tem-perature measurements from four weather stations acrossOregon (Bend, Baker, Crater Lake and the Dalles) wereused as the base climate data. We then calculated our cli-mate layer from a digital elevation model (DEM) and thedry adiabatic lapse rate of 10 �C/1000 m using AcrView.We tested the model output developed from these param-eters against our known pandora moth outbreak sites.

RESULTS

Over the 433-year period, the outbreaks appeared spatiallydispersed throughout the sample area, affecting only a few ofthe tree-ring sites at any given time (Fig. 2). The study siteswere randomly located throughout the study area, but sevenof the 13 eligible sites (two of the outbreaks only affected asingle site and were therefore not eligible for the nearestneighbour analysis) were not random (Z ¼ 1.96; a ¼ 0.05);1651 (2.34), 1676 (3.61), 1718 (4.11), 1735 (2.84), 1771(2.04), 1889 (2.72) and 1989 (6.86). Two of the outbreaks(1794 and 1836) that were randomly located were extremeevents affecting at least 75% of the sites.

The modern outbreak was well within the historicalrange of variability, although it was the fourth largestoutbreak of the 15 that were measured. The size of theoutbreaks ranged from 12.4 to 3,391.5 km2 with a mean of527.9 km2. The 1989 outbreak affected 807.9 km2

(Table 2). The histogram of the outbreak sizes describes areverse J-shaped curve with few large outbreaks and manysmaller outbreaks.

The GIS model demonstrated that, based on the host treespecies available, pandora moth could affect 43,919 km2 inOregon (Fig. 3). When we included the climate layer, wefound that climate did not greatly constrain the model, onlyrestricting the area to 40,140 km2 susceptible to outbreaks.Soils appeared to be the most limiting factor, reducing thehazard area to a region along the leeward side of the Cascademountain range. The combination of these three layersresulted in a predicted affected area totalling 12,523 km2

(Fig. 3).

DISCUSSION AND CONCLUSIONS

The modern outbreak of 1988–96 was compared with out-break history to determine if it fell within the historical rangeof variability. A pattern of change for pandora moth out-breaks is not apparent on the landscape as the modernoutbreak was within the historical range of variability andalso the fourth largest outbreak recorded. The reverseJ-shaped curve of outbreak size (few large outbreaks andmany smaller outbreaks) is frequently found in ecologicalliterature and is an expected distribution (Brown, 1995)suggesting that pandora moth has not been affected byrecent land use changes. These findings contrast sprucebudworm studies that have found an increase in severity andextent of twentieth century outbreaks (Blais, 1983; Swetnamet al., 1995). We expected that pandora moth wouldrespond to logging, which removes its primary food sourceand fragments ponderosa pine stands, in the opposite man-ner than that of the spruce budworm, resulting in smaller,

Table 1 Site characteristics

Site name Latitude, N Longitude, W Elevation (m) Tree species Soil suborder

Lookout Mountain Lower 43�45¢ 121�39¢ 1320 PIPO þ PICO CambidsPringle Falls Prescribed Fire 43�44¢ 121�39¢ 1320 PIPO þ PICO Cambids

Experimental Forest 43�43¢ 121�36¢ 1530 PIPO Orthents

Pringle Falls 43�42¢ 121�37¢ 1460 PIPO þ PICO Orthents

Surveyor Flow 43�37¢ 121�18¢ 1550 PIPO OrthentsDeschutes 43�28¢ 121�24¢ 1420 PIPO Orthents

Junction of Roads 51 and 97 43�19¢ 121 45¢ 1420 PIPO Orthents

Skookum Butte 43�14¢ 121�39¢ 1670 PIPO þ PICO OrthentsDiamond Lake 43�05¢ 121�57¢ 1510 PIPO Xerepts

Blue Jay Spring 42�55¢ 121�32¢ 1490 PIPO þ POTR Orthents

Telephone Draw 42�56¢ 121�37¢ 1550 PIPO Orthents

Telephone Draw South 42�45¢ 121�31¢ 1550 PIPO OrthentsCrater Lake 42�47¢ 122�04¢ 1370 PIPO þ ABCO Cryepts

Calimus Butte 42�38¢ 121�32¢ 2020 PIPO þ ABCO þ PILA Orthents

Species codes: ABCO, Abies concolor; PICO, Pinus contorta; PIPO, Pinus ponderosa; PILA, Pinus lambertiana; POTR, Populus tremuloides.



less-severe outbreaks. This is a much more difficult signal todiscern than the increased severity and expanse of modernspruce budworm outbreaks, with their higher mortalityrates, often as much as 80%, in heavily defoliated stands(Swetnam et al., 1995). Fragmentation by logging and theintroduction of roads may not have impacted pandora mothoutbreak dynamics because the moths are strong flyers(Massey, 1940) and are able to access the landscape at abroader scale.

We should note the difference in start dates of the modernoutbreak. Foresters were able to document the rise of themodern outbreak starting in 1988. The pattern in the wood(as reported in Table 2) runs from 1991 to 1995. The tree-ring record ended in 1995 explaining the early end date ofthat signature. The beginning date obviously does not recordthe very early emergence of the pandora moth population.The numbers of larvae have to be great enough to seriouslydamage the photosynthetic potential of the trees to berecorded in the rings. Because of this and the possibility ofnot sampling the site from which the outbreak originated,we are not able to record the very first year of the outbreaks.This suggests that, in future work, any climatic trigger thatcould be explored should come a number of years before theoutbreak is recorded in the trees.

We found that the majority of pandora moth outbreaksare significantly spatially autocorrelated, suggesting that

Figure 2 Minimum bounding polygons of the 16 pandora moth outbreaks. Each dot represents one of the 14 sites and the shaded areadesignates the area affected by each outbreak.

Table 2 The area affected by each outbreak episode. The total

study area designated the maximum possible area that could be

defoliated

Outbreak dates Area affected (km2)

1991–95 807.91967–75 13.0

1923–31 21.0

1889–03 33.4

1870–76 25.21840–53 3391.5

1803–11 1823.8

1775–83 105.31754–60 62.2

1735–42 30.9

1719–25 12.4

1677–85 71.31661–66 238.9

1652–57 39.3

1631–38 1178.4

1619–24 13.31563–78 80.8

Total study area 4413.2

Mean 527.9

Standard error 246.8Minimum 12.4

Maximum 3391.5



these outbreaks may be spreading from a local source ratherthan emergent on the landscape as a whole. We noted that itwas difficult to specify the beginning year of the outbreaksbecause it varied from one site to the next. This shows thatoutbreaks are not being triggered on all of the sites simul-taneously as it would be by favourable climate. This does notsuggest that climatic factors are not the initial trigger of thenucleus of the outbreak or important in the spread andlongevity of the outbreaks. These, hypotheses will have to betested in later work.

We found that the 1794 and 1836 outbreaks wererandomly distributed, as were the whole of our samplesites. These two outbreak events were extreme, affectingmore than 75% of the sites and probably reflected therandom locations of the sample sites. This means that forthese two events we sampled a smaller area than thefunctional area of those particular outbreaks, although, forthe majority of outbreaks, the sampling scale was appro-priate for testing the tendency of pandora moth outbreaksto be clustered.

The hazard map produced in this study agrees with Car-olin & Knopf’s (1968) published map of pandora mothoutbreak areas in Oregon and also with written accounts ofpandora moth outbreaks (Massey, 1940). As we focused onOregon and used remotely sensed images of the host treespecies, our map has greater resolution than previously

published maps. This approach examines the mechanismsthat drive the distribution of outbreaks and provides a test-able hypothesis of the areas susceptible to pandora mothoutbreaks. Future research will test this model by collectingtree rings samples located within as well as outside the areaspredicted to be susceptible to pandora moth outbreaks. Asdendrochronology will be able to determine whether pan-dora moth outbreaks have ever affected the living stand, weshould be able to quickly refine our susceptibility model.These further samples will enable us to explore the full rangeof soils, vegetation, and climate that are conducive to pan-dora moth outbreaks resulting in a more accurate model ofpandora moth hazard.

We were surprised to find that the climate layer was not amajor restraint on the range of pandora moth in our model.A scarcity of climate stations at high elevations hinders thedevelopment of a true climate layer based solely on meas-urements. In future work, we may try to refine our climatelayer by using meteorological data from more stations orclimate divisions throughout the area. Climate division dataextrapolates over the entire divisional area, which is brokendown by major topographic barriers and air mass bound-aries. This may be the best source of data for future mod-elling efforts.

One such exploratory analysis has been completed whichsupports this hazard model. Speer et al. (1997) sampled near

Vegetation susceptible to Pandora Moth

Climate and vegetation susceptible to Pandora Moth

Soils, climate and vegetation susceptible to Pandora Moth

0 30 60 Kilometres

Figure 3 Hazard map of the area susceptible

to pandora moth in Oregon. This map was

compiled by overlaying vegetation, climate,

and soil layers to determine the area thatwould be favoured by pandora moth.



Sisters, Oregon (44.37 N, 121.55 W), c. 100 km north ofour northernmost sites but within our predicted area ofpandora moth hazard. Nine pandora moth outbreaks wererecorded extending back to AD 1610. Foresters have notpreviously reported the occurrence of pandora moth out-breaks on this site. Speer et al. (1997) found that the mostrecent outbreak ended in AD 1895, explaining why pandoramoth outbreaks had not been recorded this far north.

Spatial analysis of past insect outbreaks could enableforest managers to better understand the outbreak charac-teristics of an endemic insect in the systems they are man-aging. This insect has been present for the entire 433 years ofthis record and the 1988 outbreak was within the historicalrange of variability. We were able to document that theseoutbreaks spread from a nucleus and do not affect all sitessimultaneously. Future work must be carried out to examinethe mechanism driving the spread of pandora mothoutbreaks. The GIS hazard model appears to be a usefuldeterminant of areas that are potentially susceptible topandora moth outbreaks and can be used as a managementtool to inform local forest managers of the potential offuture pandora moth outbreaks in their forest units. It canalso be used as a testable model of where pandora moth hasoccurred in the past. This model will help drive futuresampling and those samples will help to refine this hazardmodel. This paper has laid the groundwork for the use ofGIS to model insect outbreak hazard. Now this work can beexpanded to examine the entire range of pandora moth inthe western United States and applied to other insects thatare dependent upon specific environmental conditions.

ACKNOWLEDGMENTS

We would like to thank Jay Gatrell and two anonymousreferees for their comments on earlier drafts of this manu-script.

REFERENCES

Baker, W.L. (1995) Long-term response of disturbance land-scapes to human intervention and global change. LandscapeEcology, 10, 143–159.

Baker, W.L. & Kipfmueller, K.F. (2001) Spatial ecology of Pre-Euro-American fires in a southern rocky mountain subalpineforest landscape. Professional Geographer, 53, 248–262.

Blais J.R. (1983) Trends in the frequency, extent, and severity ofspruce budworm outbreaks in eastern Canada. CanadianJournal of Forest Research, 13, 539–547.

Brown, J.H. (1995) Macroecology. The University of ChicagoPress, Chicago, IL, 269 pp.

Carolin, V.M. & Knopf, J.A.E. (1968) The pandora moth. USDepartment of Agriculture, Forest Service, Washington, DC,Forest Pest Leaflet, Vol. 114, 7 pp.

Earickson, R.J. and Harlin, J.M. (1994) Geographic measure-ment and quantitative analysis. Macmillan College PublishingCompany, New York, 350 pp.

Eeley, H.A.C., Lawes, M.J. & Piper, S.E. (1999) The influenceof climate change on the distribution of indigenous forest in

KwaZulu-Natal, South Africa. Journal of Biogeography, 26,595–617.

Fitzgerald, S.A. (1992) Controlling pandora moth in centralOregon forests and landscapes. Oregon State UniversityExtension Service, Corvallis EC 1383, 4 pp.

Franklin, J.F. & Dyrness, C.T. (1988) Natural vegetation ofOregon and Washington. Oregon State University Press,Corvallis, 452 pp.

Furniss, R.L. & Carolin, V.M. (1977) Western forest insects. USDepartment of Agriculture, Forest Service, 654 pp.

Jensen, R.R. (2002) Spatial and temporal leaf area indexdynamics in a north central Florida, USA preserve. GeocartoInternational, 17, 45–52.

Jensen, R.R. and Carson, A.R. (2001) Longleaf Pine/TurkeyOak Sandhill Loss in a North Central Florida Preserve, 1972–1997. Southeastern Geographer, 41, 306–311.

Massey, C.L. (1940) The pandora moth (Coloradia pandoraBlake), a defoliator of lodgepole pine in Colorado. MAThesis, Duke University, Duke, 33 pp.

Morgan, P., Aplet, G.H., Haufler, J.B., Humphries, H.C.,Margaret, M.M. & Wilson, W.D. (1994) Historical rangeof variability: a useful tool for evaluating ecosystem change.Assessing forest ecosystem health in the inland west.Proceedings of the American forests scientific workshop(ed. by R.L. Sampson and D.L. Adams), pp. 87–111. TheHawthorn Press Inc., New York.

Mutch, R.W., Arno, S.F. & Brown, J.K. (1993) Forest health inthe blue mountains: a management strategy for fire-adaptedecosystems. US Department of Agriculture, Forest Service,Pacific Northwest Research Station, Portland, OR. GeneralTechnical Report PNW-GTR-310, 14 pp.

Patterson, J.E. (1929) The pandora moth, a periodic pest of thewestern pine forests. US Department of Agriculture, ForestService. Technical Bulletin 137, 20 pp.

Schmid, J.M. & Bennett, D.D. (1988) The North Kaibab pandoramoth outbreak, 1978–1984. US Department of Agriculture,Forest Service, Rocky Mountain Forest and Range Experimen-tal Station, Fort Collins, CO. Research Note RM-153, 20 pp.

Speer, J.H. (1997) A dendrochronological record of pandoramoth (Coloradia pandora, Blake) outbreaks in central Ore-gon. MS Thesis, The University of Arizona, Arizona, 159 pp.

Speer, J.H., Cushman, L., Negron, J., Davi, N., Schauer, A. &Darbyshire, R. (1997) Pandora moth outbreak chronology forthe eastern cascades, Deschutes National Forest, Oregon. Theeighth annual North American dendroecological fieldweek:final report, July 1997 (ed. by P.M. Brown and P.J. Krusic).H.J. Andrews Research Station, Oregon, 4 pp.

Speer, J.H., Swetnam, T.W., Wickman, B.E. & Youngblood, A.(2001) Changes in pandora moth outbreak dynamics duringthe past 622 years. Ecology, 82, 679–697.

Swetnam, T.W., Wickman, B., Paul, G.H. & Baisan, C.H.(1995) Historical patterns of western spruce budworm andDouglas-fir tussock moth outbreaks in the Northern BlueMountains, Oregon. US Department of Agriculture, ForestService, Pacific Northwest Research Station. Research PaperPNW-RP 484, 27 pp.

Wickham, J.D., O’Neill, R.V. & Jones, K.B. (2000) Forestfragmentation as an economic indicator. Landscape Ecology,15, 171–179.



BIOSKETCHES

Jim Speer is an Assistant Professor of Geography and Geology. He is interested in environmental reconstruction using treerings and has worked on projects reconstructing insect outbreaks, fire history, climate, ice storms, and acorn production in oaktrees.

Ryan Jensen is an Assistant Professor of Geography. He is interested in longleaf pine/turkey oak sandhill fire ecology andhabitat loss, tropical deforestation, and urban forestry.



Date post:	13-Nov-2023
Category:	Documents
Upload:	indstate
View:	1 times
Download:	0 times

A hazards approach towards modelling pandora moth risk

Documents